INDEX

Explanations

terms related to laws, specifically focusing on prohibition

oai_token-act-pair · gpt-3.5-turbo

references to prohibition laws

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Fault

-0.65

events

-0.65

Tur

-0.65

Tur

-0.65

PE

-0.64

ource

-0.63

eon

-0.63

 Oval

-0.62

rics

-0.61

)</

-0.59

POSITIVE LOGITS

 prohibition

1.20

 regimes

0.87

ategory

0.85

 prohibited

0.85

 prohibit

0.84

 outlaw

0.84

 preclude

0.84

 prohibiting

0.83

 restriction

0.82

alist

0.82

Activations Density 0.015%

terms related to laws, specifically focusing on prohibition

references to prohibition laws

No Comments

No Known Activations